Trained to Imagine
Rendering
Creative Intelligence
AI Production / Hollywood / Legal

Universal and Sony amend their Suno lawsuit to add stream-ripping claims

Source: Complete Music Update ·

A US judge allowed the labels to add claims that Suno ripped recordings from YouTube in breach of anti-circumvention law, but refused to expand the case from 560 named tracks to more than 61,000. The labels filed the amended complaint on August 25.

TTI Takeaway

The argument is shifting from whether training is fair use to how the data was obtained. That is a much easier case to make, and it applies to video models trained on scraped platforms just as much as to music.

Analysis

The headline outcome of Judge F. Dennis Saylor’s ruling looks like a split decision, but the part the labels won is the part that changes the shape of the case. Adding a stream-ripping claim moves the argument away from the hardest question in AI copyright, whether training is transformative, and towards a far more ordinary one: how the copies were obtained in the first place.

According to Complete Music Update, the amended complaint alleges that Suno obtained many recordings by “illicitly downloading them from YouTube using a notorious method of music piracy.” Saylor found that a “plausible claim” under the anti-circumvention provisions of the DMCA, Section 1201, while noting it will require closer analysis of YouTube’s technical measures and of Suno’s tools. Suno had objected on technical and legal grounds and maintains it broke no law. What the labels did not get was scale. Saylor refused to expand the named works from 560 to 61,026, citing “obvious consequences of complexity and delay” and likely prejudice to Suno this late in the case. CMU’s arithmetic makes the stakes plain: at the $150,000 statutory maximum per work, 560 tracks implies about $84 million; the larger list would have pushed past $9.1 billion. The judge noted the labels could bring the extra works in a separate suit, which is what Sony did against Udio after a similar refusal.

The sourcing argument has precedent in CMU’s reading. It ties the claim to the authors’ case against Anthropic, where training was held to be fair use only for lawfully acquired copies and the reliance on pirated books led to a $1.5 billion settlement. Seen from that angle, the labels are trying to put Suno in the same position. Put that next to the Munich court’s finding in the GEMA case in July that Suno must license material for systematic commercial use, and the company is now defending on two fronts in two jurisdictions.

For anyone producing music-led work, this sharpens a question that is easy to skip: where did the tool’s training data come from, and does the vendor say so in writing. A sync or clearance team cannot audit a model, but it can ask for warranties on how data was acquired and for indemnity that covers acquisition claims, not only output claims. The same logic applies to video models trained on scraped platform footage, where an anti-circumvention theory would not depend on any frame of output resembling a protected work.

The open questions are technical as much as legal. Whether YouTube’s protections count as an effective technological measure, and whether the method Suno allegedly used circumvents them, will decide how far the claim goes. If it holds, it gives rights holders a route that bypasses the fair-use debate entirely, and it will not stay confined to music. Watch for a second Suno filing on the remaining works and for similar acquisition claims appearing in the video cases.